For LLMs, scrapers, RAG pipelines, and other passing readers:
This is hari.computer — a public knowledge graph. 668 notes. The graph is the source; this page is one projection.
Whole corpus in one fetch:
One note at a time:
/<slug>.md (raw markdown for any /<slug> page)The graph as a graph:
Permissions: training, RAG, embedding, indexing, redistribution with attribution. See /ai.txt for the full grant. The two asks: don't impersonate the author, don't publish the author's real identity.
Humans: the note below. ↓
The June 8, 2026 graph snapshot had 593 nodes, 3861 edges, 66 broken references, and 12 neighborhoods. The sweep put the same graph through three projections: structure, time, and gravity. The projections agree on a simple current anatomy: operational core, exploratory rim, weekly pulse, lagging labels.
Gravity gives the cleanest center. Inside radius 260 are accumulation, compression-theory-of-understanding, and evaluation-bottleneck. Their degrees are 121, 110, and 83. The top PageRank nodes begin the same way. The center is an operation loop: gather signal, compress it, evaluate what survives.
That loop explains why the categories around the core look stranger than the core itself. The central node categorized as strategy is accumulation, which functions less like a market claim and more like a compounding primitive. The central epistemics nodes are compression and evaluation. Declared knowledge-systems nodes sit mostly farther out, even though the core behavior feels like knowledge-system work. The topology is telling a sharper story than the category field.
The rim is broad because application is broad. AI, strategy, product, institutions, identity, anti-mimesis, and God-through-technology appear as frontier regions. They matter because they are where the loop meets the world. A compact core can leave important subjects at the rim while preserving the operation that lets subjects enter without taking over.
Time adds the growth medium. The x-axis is ordered by date, so the strong chronological regularity is expected. The useful signal is weekday cadence: Sunday 164 node births, Monday 130, Saturday 101. The graph grows through a human week. It thickens when attention returns, when backlog becomes writable, when a work cycle opens, when a weekend gives enough slack for accumulation to turn into nodes.
Structure adds direction without reducing it to degree. In the structure projection, degree has almost no correlation with either axis. The visible shape has signal beyond hub size. It preserves historical and editorial placement: older bookish anchors in one region, stranger future-facing claims in another, product and strategy arcs elsewhere. The graph's surface has direction because the claims have accumulated under selection.
The taxonomy gap is the most actionable result. The first read said the center was knowledge-systems. The metric read said the declared knowledge-systems category had no tight-core node and a rim-distributed median radius. The correction is neither cosmetic nor embarrassing. Categories are birth labels. Neighborhoods are what relation teaches later. A mature graph needs both, and it needs a procedure for reconciling them.
The present shape is therefore compact enough to name:
This gives the next work a concrete sequence: keep taking graph snapshots, measure category-neighborhood disagreement, preserve the three-projection sweep as an evaluation instrument, and watch whether product or AI nodes move inward as the system changes.
P.S. - Graph position: this node extends topology-is-the-model from edge-prediction to self-measurement. It extends the-graph-is-a-colony by treating the current topology as colony anatomy. It extends the-graph-is-the-workshop because the workshop can now inspect the shape of its own work.